Many video encoders support an inter-image coding mode (interframe coding hereafter interceding), in which the current image is coded with respect to one or more preceding images.
Each image can also be coded without reference to the others, that coding is called intraframe coding (hereafter intracoding). For a given transmission rate, intracoding gives poorer video quality than does interceding since it does not take advantage of the temporal correlations between the successive images of the video sequence.
Currently, a portion of video sequence has its first image intracoded then the following images are intercoded.
A problem with interceding is its behaviour in the presence of transmission error or packet losses on the transmission channel. The degradation or the loss of an image propagates over the following images until a new intracoded image appears.
It is common that the transmission mode of the signal generates total or partial losses of certain images. From such losses there results, for example, the loss or the late arrival of certain data packets when the transmission takes place on a packet network with no guarantees of delivery such as the IP (Internet Protocol) network. Losses can also result from errors introduced by the transmission channel beyond the correction capabilities of the correcting codes used.
In an environment subject to various signal losses, it is necessary to provide mechanisms for improving the quality of the image at the level of the decoder. One of these mechanisms is the use of a feedback channel, from the decoder to the coder, upon which the decoder informs the coder that it has lost all of part of certain images. In certain cases, it is the well-reconstructed images that the decoder indicates to the coder on the feedback channel so that the coder can derive from this which images have possibly been lost.
In response to each of these error messages, most coders systematically return intracoded correction images, i.e. without reference to the preceding images, which makes it possible to refresh the display and to correct errors due to transmission losses.
The correction images are transmitted in response messages during specific transmissions or by insertion in the stream of the video sequence in replacement of other images.
Such a use of the feedback channel has significant disadvantages when several error messages are transmitted on the feedback channel. The present principle results in a response being given to each of these messages. For example, if an image is lost, the decoder is likely to send an error message on the feedback channel for each of the following interceded images with a reference to the lost image. In this case, the encoder is made to transmit several response messages each comprising an intracoded correction image. This results in a reduced restoration quality and a significant occupation of bandwidth.
Similarly, in certain multi-broadcast embodiments, such as conferencing bridge systems or others in which several decoders receive the same information, it happens that each of the decoders transmits an error message on a feedback channel relating to a same lost image and that the encoder responds successively to each of these messages retransmitting the same correction image several times.
Thus, the use made of feedback channels is not optimum in particular when several messages are transmitted on the feedback channel.
One of the advantages of the present invention is to improve the use of the feedback channel.
The invention thus proposes a method for broadcasting images from a video sequence, the method comprising the following steps, at the level of broadcasting equipment:                including images in an output stream for transmission to at least one restoration equipment;        receiving feedback from said restoration equipment of error messages relating to the restoration of the images from the video sequence;        analyzing the error messages in order to identify images or parts of images not restored or badly restored;        determining response messages including images or image of correction parts,characterized in that the method furthermore comprises:        determining a level of seriousness for each error message according to the analysis performed; and        deciding to return a response message relating to an error message according to the level of seriousness of that error message.        
Consequently, the method of the invention comprises an analysis of the error messages and the quantification of the seriousness of these messages in order to be able to decide if it is appropriate to respond to them. This quantification makes it possible to optimise the use of the feedback channel by the optimisation of the response messages.
In a particular embodiment, said analysing comprises determining at least one factor originating from the following group:                a restoration quality level taking account of the error;        an evaluation of the importance of the content of images or image parts which are not restored or badly restored;        decoding parameters;        an evaluation of the transmission conditions; and        an estimation of the time between the current image and the last image transmitted without error,said factor being used for determining the level of seriousness associated with an error message.        
Advantageously, said level of seriousness of at least one error message is a maximum waiting time before the transmission of a response message. This makes it possible to ensure the transmission of a response message for certain error messages.
As a variant, the method comprises determining an emergency level from the levels of seriousness associated with the error messages received since the transmission of the last response message and said deciding is carried out according to this emergency level. This embodiment has the advantage of determining an emergency quantifier from several error messages.
Advantageously, said determining response messages comprises grouping response messages. This makes it possible to reduce the number of response messages with respect to the number of error messages.
In a particular embodiment, said analysing comprises identifying a last image transmitted without error and said determining a response message comprises encoding a correction image with respect to said last image transmitted without error. This embodiment makes it possible to avoid the transmission of an intracoded image in order to improve the quality of the restoration.
Alternatively, said including images in a stream comprises formatting several input streams in order to form said output stream, said analysing the error messages comprises identifying images which are not restored or badly restored in said input streams and said determining a response message comprises modifying correction images originating from said input streams in order to make them conform with the format of the output stream.
Moreover, the invention also relates to a computer program or broadcasting equipment implementing the method previously referred to as well as to an image broadcasting system using such broadcasting equipment.